15 curated agentic skills for AI-assisted development workflows. Each skill is a self-contained prompt that teaches a coding agent a specific capability. Works with Claude Code and OpenAI Codex.
| Skill | Description |
|---|---|
| code-review-rust | Rust review (error handling, safety, idioms) |
| code-review-ts | TypeScript review (type safety, idioms) |
| differential-review | Security-focused review for PRs and diffs |
| Skill | Description |
|---|---|
| function-analyzer | Per-function deep analysis for audit context |
| semgrep | Semgrep static analysis with parallel scan/triage |
| spec-compliance-checker | Spec-to-code compliance analysis |
| Skill | Description |
|---|---|
| tdd | Test-driven development (RED, GREEN, REFACTOR) |
| Skill | Description |
|---|---|
| dogfood | Use the tools you build. No manual workarounds. |
| model-router | Route work to the cheapest capable model tier |
| rick-rubin | Scope discipline and simplicity |
| Skill | Description |
|---|---|
| paul-graham | Direct prose, structural editing, README quality |
| Skill | Description |
|---|---|
| refactor-suggestions | Suggest refactors for modified code |
| repo-hygiene | Pre-work checks, documentation updates, cleanup |
| Skill | Description |
|---|---|
| workflow | Structured cycle: brainstorm, plan, work, review, compound |
| Skill | Description |
|---|---|
| skill-maker | Create new agentic skills |
Copy a skill into your project:
cp -r semgrep /path/to/project/.claude/skills/Reference it in your project's CLAUDE.md:
## Installed Skills
- @.claude/skills/semgrep/SKILL.mdEvery skill includes an agents/openai.yaml config. Point your Codex agent at the yaml file for the skill you want to use.
./validate.sh # all skills
./validate.sh semgrep # single skillRequires the hooks repo at ../hooks.
MIT